update

deepseek_vl2/serve/inference.py

@@ -39,7 +39,7 @@ def load_model(model_path, dtype=torch.bfloat16):
     tokenizer = vl_chat_processor.tokenizer
 
     vl_gpt: DeepseekVLV2ForCausalLM = AutoModelForCausalLM.from_pretrained(
-        model_path, trust_remote_code=True, torch_dtype=dtype
+        model_path, force_download=False, trust_remote_code=True, torch_dtype=dtype
     )
     if torch.cuda.is_available():
         vl_gpt = vl_gpt.cuda()